Zero-density sets containing integers with at most two prime factors (Q5960979)

Let \({\mathcal D}\) denote a proper subset of \(\{0,1,\dots, r-1\}\) with \(t\) elements \((2\leq t< r)\), and \(W_{\mathcal D}(X)\) denote the set of integers \(n\) \((0\leq n< X)\) of the form \(n= \sum_{0\leq j< s}a_jr^j\) with each \(a_j\in{\mathcal D}\). The authors continue their investigations into the number of elements of \(W_{\mathcal D}(X)\) with few prime factors. In [J. Number Theory 81, 270-291 (2000; Zbl 0957.11039)], the proved that there exists \(k=k(r)\in \mathbb{N}\) such that \[ |\{n\in W_{\mathcal D}(X): \Omega(n)\leq k\}|\gg_{r,{\mathcal D}}|W{\mathcal D}(X)|/\log X, \tag{*} \] where \(\Omega(n)\) is the number of prime factors of \(n\). They also determined conditions on \(t\) (of the form \(t> r^\alpha\), \(\alpha\geq\frac 12\)) such that (*) holds with \(k=3,4,5\), and considered the special case \({\mathcal D}= \{0,1\}\) for \(r\geq 3\). Improvements of some of these results are established in the current paper. When \({\mathcal D}= \{0,1,\dots, t-1\}\), the authors prove that (*) holds with \(k=2\) when \(t=r-1\) and \(r\geq 5\), and when \(t> r^\alpha\) with \(\alpha=0.512\) and \(r\geq r_0(\alpha)\). When \({\mathcal D}= \{0,1\}\), \(r=3\), they indicate that their method gives (*) with \(k=3\), an improvement of their earlier result. The standard method used to establish results of this type is to apply a sieve, in the case a weighted sieve discussed by \textit{G. Greaves} in [Number theory, trace formulas and discrete groups (Oslo, 1987), 289-308 (1989; Zbl 0682.10038)]. A crucial step in the application is embodied in Theorem 2, which is too complicated to state here. The final theorem of this paper concerns a generalization \(\Lambda_k\) of the von Mangoldt function, given by \(\Lambda_k= \mu*(\log)^k\) where \(\mu\) is the Möbius function. It is known that \(\Lambda_k\geq 0\) and that \(\Lambda_k(n)=0\) when \(n\) has more than \(k\) distinct prime factors (not, as stated, when \(\Omega(n)> k\), unless \(\mu^2(n)= 1\)). When \({\mathcal D}= \{0,1,\dots, t-1\}\), the authors use their sieve methods to establish, under certain conditions on \(t\) and \(X\), an asymptotic formula for the sum \(\sum_{n\in W_{\mathcal D}(X)} \mu^2(n) \Lambda_k(n)\).